Prompt library

37 questions to ask your AI about your own body

Copy any of these into Claude, Cursor or VS Code once FitnessPeak is connected. The interesting ones aren't the questions your app already answers — they're the ones that need two sources joined together.

TL;DR. 37 prompts across 6 themes: training decisions, sleep, strength progression, training load, body composition, and the cross-source questions no single app can answer. Each is listed with the reason it needs FitnessPeak rather than the WHOOP, Hevy or Withings app. Works with any MCP client. Setup takes about two minutes.

Jump to: Should I train today? · Why am I tired? · Am I actually getting stronger? · How hard am I actually training? · What has actually changed? · The ones no single app can answer

01 · 6 prompts

Should I train today?

The question every strap owner asks every morning. The app gives a number; your assistant can give a decision, because it also knows what you have planned.

AskWhy it needs FitnessPeak
Should I do my scheduled hard session today, given my recovery?Weighs recovery against what's actually in your calendar
My recovery is 41%. What did I do the last three times it was this low?Needs history, not today's number
Is it better to train hard today or tomorrow, based on the trend?A forecast from a trend, not a snapshot
How many days this month was I green, yellow and red?Aggregation across an arbitrary window
Has my resting heart rate changed since I stopped drinking?Correlates a habit against a metric
Am I more recovered on days I skip evening screens?Joins a behaviour you mention against WHOOP data
02 · 7 prompts

Why am I tired?

Sleep data is where the gap between a chart and an answer is widest. Stage breakdowns only become useful once something correlates them with the rest of your life.

AskWhy it needs FitnessPeak
I slept 8 hours and still feel wrecked. Why?Separates time in bed from time actually asleep
Which nights had the least deep sleep, and what was my strain the day before?Joins sleep against training load
Compare my sleep stages this month against last month.Two arbitrary windows in one answer
Has my sleep efficiency improved since I moved my bedtime earlier?Before-and-after on a change you made
Include naps — how much total sleep am I really getting?A parameter the app doesn't expose
Which nights did I wake up most, and were they after late sessions?Cross-references wake-ups with workout times
What's my longest streak of 7+ hour nights this year?A streak calculation over a year of data
03 · 7 prompts

Am I actually getting stronger?

Hevy knows what you lifted. WHOOP knows what it cost you. Neither app can answer a question that needs both.

AskWhy it needs FitnessPeak
Is my squat going up? Show me every session this year.Per-exercise progression across sessions
What was my best set of bench press, and when?Estimated 1RM across your whole history
How much total volume did I lift this month versus last?Volume is Σ weight × reps, computed not stored
Which lifts have stalled for more than six weeks?Needs a trend per exercise, not a list
Do I lift more on high-recovery days?Joins Hevy volume against WHOOP recovery
What's my most neglected muscle group this quarter?Aggregates the exercise catalogue by muscle
Am I training legs less often since I started running?Two sources, one behavioural question
04 · 6 prompts

How hard am I actually training?

Day strain counts everything, not just what you remembered to log. That's the number to use for weekly load.

AskWhy it needs FitnessPeak
What's my weekly training load been for the last two months?Whole-day strain, aggregated by week
Which days had high strain but no logged workout?Finds load you didn't record
Am I absorbing my current training block or overreaching?Falling recovery + rising RHR across weeks
Compare this training block to the same phase of my last one.Two arbitrary historical windows
What was my hardest session this year, by strain?A max across everything you've done
Should I move my down week forward?A recommendation from your own response history
05 · 6 prompts

What has actually changed?

Withings measures the slow-moving numbers. On their own they're noise; against training and recovery they're a story.

AskWhy it needs FitnessPeak
Is my weight trending up or down over six months?A trend line, not a scatter of readings
Has my body fat percentage moved while my weight stayed flat?Two measures compared over one window
Did my muscle mass change during my last strength block?Joins Withings against Hevy dates
What's my blood pressure history, and is it drifting?Sporadic readings across years
Show my resting heart rate and weight on the same timeline.Two sources, one comparison
Am I lighter on mornings after high-strain days?A same-day join across two devices
06 · 5 prompts

The ones no single app can answer

These are the reason FitnessPeak exists. Each needs two or three sources reconciled before a model sees anything.

AskWhy it needs FitnessPeak
Write my weekly training review: load, recovery, sleep and what to change.All three sources, written as prose
Which type of training costs me the most recovery — lifting or running?Hevy exercise type against WHOOP recovery
Correlate my sleep with the next day's lifting volume.A computed statistic across two sources
Build me a training plan for next week based on this week.Reads your data, writes a plan
What single change would most improve my recovery?Ranks candidate causes against your history

Getting better answers

  • Give it context it can't fetch. Your training plan, your race date, how you actually felt. The data is only half the picture.
  • Ask follow-ups. The data stays in the conversation, so "what about the month before?" costs one sentence instead of another query.
  • Ask for a format. "As a table", "as a message to my coach", "in three bullets" — the model is doing the writing, so tell it how.
  • Push back. If a number looks wrong, say so. It can re-query a different range rather than defend itself.

Questions

Do I have to phrase questions exactly like this?
No. These are starting points, not commands. Your assistant reads the tool descriptions and works out which data it needs, so plain language is enough — 'how did last week go?' works.
Do I need all three sources connected?
No. Every question is tagged with the sources it uses. WHOOP alone answers the recovery, sleep and strain sections; the cross-source questions need two or more.
Can my assistant get this wrong?
Yes. It is a language model reading tables. FitnessPeak returns the data and tells it that missing rows mean the strap was not worn rather than a zero, which prevents the most common class of confident error — but always sanity-check a surprising claim against the source app.
Will it remember my answers between conversations?
That depends on your AI client, not on FitnessPeak. We return fresh data on every call; whether the conversation persists is your assistant's memory feature.
Is this medical advice?
No. These are fitness metrics and an assistant's interpretation of them. Talk to a clinician before acting on anything health-related.

Ask these about your own data

14-day free trial. Card required so there's no second signup at the end; cancel before day 14 and you're charged nothing.